Refreshing Thoughts on DRAM : Power Saving vs . Data Integrity
Refreshing Thoughts on DRAM : Power Saving vs . Data Integrity
复制标题
关于 DRAM 的新想法：节能与节能
DOI：
--
复制
发表时间：
2014
期刊：
影响因子：
--
通讯作者：
Kevin Fu
中科院分区：
文献类型：
--
作者：
Amir Rahmati;Matthew Hicks;Daniel E. Holcomb;Kevin Fu
为了阻止DRAM中由于刷新而导致的功耗和吞吐量开销增加的趋势，研究人员正在探索微调刷新率的方法。刷新管理建议的范围从温度感知刷新到基于易失性对数据单元进行分区，每个分区都有自己的刷新率。这方面的一个障碍是缺乏对当前提案中使用的评价设置的准确描述。不完整的描述使得很难重现结果和比较方法。因此，研究人员通常使用数学模型来评估他们的方法，这些模型是根据来自不同平台的实验结果得出的。为了帮助评估现有的和未来的调整DRAM刷新以节省功率的方法，本文提供了一个可重复的DRAM研究平台。在这个平台上，我们重新运行了该领域以前论文中经常引用的实验。我们的实验结果通过显示假设和实验结果对温度和刷新持续时间等经常被忽略的参数的敏感性，突出了可重复平台的必要性。
To head-off the trend of increasing power consumption and throughput overheads due to refresh in DRAM, researchers are exploring ways to fine-tune refresh rate. Refresh management proposals range from temperature-aware refresh to partitioning data cells based on volatility, with each partition having its own refresh rate. One hurdle in this area is the lack of precise description of the evaluation setups used in current proposals. The incomplete description makes it difficult to reproduce reproduce results and compare approaches. Thus, it is common for researchers to evaluate their approach using mathematical models derived on the experimental results from a disparate set of platforms. To aide in the evaluation of existing and future approaches for tuning DRAM refresh to save power, this paper provides a reproducible DRAM research platform. On this platform, we re-run commonly cited experiments from previous papers in the area. Our experimental results highlight the necessity of a reproducible platform by showing how both assumptions and experimental outcomes are sensitive to often omitted parameters like temperature and refresh duration.